Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc

Pull two sparc fixes from David Miller: 1) Fix boots with gcc-4.9 compiled sparc64 kernels. 2) Add missing __get_user_pages_fast() on sparc64 to fix hangs on futexes used in transparent hugepage areas. It's really idiotic to have a weak symbolled fallback that just returns zero, and causes this kind of bug. There should be no backup implementation and the link should fail if the architecture fails to provide __get_user_pages_fast() and supports transparent hugepages. * git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc: sparc64: Implement __get_user_pages_fast(). sparc64: Fix register corruption in top-most kernel stack frame during boot.

Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc
Pull two sparc fixes from David Miller: 1) Fix boots with gcc-4.9 compiled sparc64 kernels. 2) Add missing __get_user_pages_fast() on sparc64 to fix hangs on futexes used in transparent hugepage areas. It's really idiotic to have a weak symbolled fallback that just returns zero, and causes this kind of bug. There should be no backup implementation and the link should fail if the architecture fails to provide __get_user_pages_fast() and supports transparent hugepages. * git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc: sparc64: Implement __get_user_pages_fast(). sparc64: Fix register corruption in top-most kernel stack frame during boot.
Linus Torvalds
2 parents 96971e9aa9 06090e8ed8
Showing 11 changed files Side-by-side Diff
arch/sparc/include/asm/oplib_64.h
arch/sparc/include/asm/setup.h
arch/sparc/kernel/entry.h
arch/sparc/kernel/head_64.S
arch/sparc/kernel/hvtramp.S
arch/sparc/kernel/setup_64.c
arch/sparc/kernel/trampoline_64.S
arch/sparc/mm/gup.c
arch/sparc/prom/cif.S
arch/sparc/prom/init_64.c
arch/sparc/prom/p1275.c
@@ -62,7 +62,8 @@
 /* You must call prom_init() before using any of the library services,
  * preferably as early as possible.  Pass it the romvec pointer.
  */
-void prom_init(void *cif_handler, void *cif_stack);
+void prom_init(void *cif_handler);
+void prom_init_report(void);
  
 /* Boot argument acquisition, returns the boot command line string. */
 char *prom_getbootargs(void);
@@ -48,6 +48,8 @@
 #endif
  
 #ifdef CONFIG_SPARC64
+void __init start_early_boot(void);
+
 /* unaligned_64.c */
 int handle_ldf_stq(u32 insn, struct pt_regs *regs);
 void handle_ld_nf(u32 insn, struct pt_regs *regs);
@@ -65,13 +65,10 @@
 extern struct pause_patch_entry __pause_3insn_patch,
 	__pause_3insn_patch_end;
  
-void __init per_cpu_patch(void);
 void sun4v_patch_1insn_range(struct sun4v_1insn_patch_entry *,
 			     struct sun4v_1insn_patch_entry *);
 void sun4v_patch_2insn_range(struct sun4v_2insn_patch_entry *,
 			     struct sun4v_2insn_patch_entry *);
-void __init sun4v_patch(void);
-void __init boot_cpu_id_too_large(int cpu);
 extern unsigned int dcache_parity_tl1_occurred;
 extern unsigned int icache_parity_tl1_occurred;
  
@@ -672,14 +672,12 @@
 	sethi	%hi(init_thread_union), %g6
 	or	%g6, %lo(init_thread_union), %g6
 	ldx	[%g6 + TI_TASK], %g4
-	mov	%sp, %l6
  
 	wr	%g0, ASI_P, %asi
 	mov	1, %g1
 	sllx	%g1, THREAD_SHIFT, %g1
 	sub	%g1, (STACKFRAME_SZ + STACK_BIAS), %g1
 	add	%g6, %g1, %sp
-	mov	0, %fp
  
 	/* Set per-cpu pointer initially to zero, this makes
 	 * the boot-cpu use the in-kernel-image per-cpu areas
  
  
@@ -706,44 +704,14 @@
 	 nop
 #endif
  
-	mov	%l6, %o1			! OpenPROM stack
 	call	prom_init
 	 mov	%l7, %o0			! OpenPROM cif handler
  
-	/* Initialize current_thread_info()->cpu as early as possible.
-	 * In order to do that accurately we have to patch up the get_cpuid()
-	 * assembler sequences.  And that, in turn, requires that we know
-	 * if we are on a Starfire box or not.  While we're here, patch up
-	 * the sun4v sequences as well.
+	/* To create a one-register-window buffer between the kernel's
+	 * initial stack and the last stack frame we use from the firmware,
+	 * do the rest of the boot from a C helper function.
 	 */
-	call	check_if_starfire
-	 nop
-	call	per_cpu_patch
-	 nop
-	call	sun4v_patch
-	 nop
-
-#ifdef CONFIG_SMP
-	call	hard_smp_processor_id
-	 nop
-	cmp	%o0, NR_CPUS
-	blu,pt	%xcc, 1f
-	 nop
-	call	boot_cpu_id_too_large
-	 nop
-	/* Not reached... */
-
-1:
-#else
-	mov	0, %o0
-#endif
-	sth	%o0, [%g6 + TI_CPU]
-
-	call	prom_init_report
-	 nop
-
-	/* Off we go.... */
-	call	start_kernel
+	call	start_early_boot
 	 nop
 	/* Not reached... */
  
@@ -109,7 +109,6 @@
 	sllx		%g5, THREAD_SHIFT, %g5
 	sub		%g5, (STACKFRAME_SZ + STACK_BIAS), %g5
 	add		%g6, %g5, %sp
-	mov		0, %fp
  
 	call		init_irqwork_curcpu
 	 nop
@@ -30,6 +30,7 @@
 #include <linux/cpu.h>
 #include <linux/initrd.h>
 #include <linux/module.h>
+#include <linux/start_kernel.h>
  
 #include <asm/io.h>
 #include <asm/processor.h>
@@ -162,7 +163,7 @@
  
 static struct pt_regs fake_swapper_regs = { { 0, }, 0, 0, 0, 0 };
  
-void __init per_cpu_patch(void)
+static void __init per_cpu_patch(void)
 {
 	struct cpuid_patch_entry *p;
 	unsigned long ver;
@@ -254,7 +255,7 @@
 	}
 }
  
-void __init sun4v_patch(void)
+static void __init sun4v_patch(void)
 {
 	extern void sun4v_hvapi_init(void);
  
  
  
@@ -323,14 +324,25 @@
 	}
 }
  
-#ifdef CONFIG_SMP
-void __init boot_cpu_id_too_large(int cpu)
+void __init start_early_boot(void)
 {
-	prom_printf("Serious problem, boot cpu id (%d) >= NR_CPUS (%d)\n",
-		    cpu, NR_CPUS);
-	prom_halt();
+	int cpu;
+
+	check_if_starfire();
+	per_cpu_patch();
+	sun4v_patch();
+
+	cpu = hard_smp_processor_id();
+	if (cpu >= NR_CPUS) {
+		prom_printf("Serious problem, boot cpu id (%d) >= NR_CPUS (%d)\n",
+			    cpu, NR_CPUS);
+		prom_halt();
+	}
+	current_thread_info()->cpu = cpu;
+
+	prom_init_report();
+	start_kernel();
 }
-#endif
  
 /* On Ultra, we support all of the v8 capabilities. */
 unsigned long sparc64_elf_hwcap = (HWCAP_SPARC_FLUSH | HWCAP_SPARC_STBAR |
@@ -109,10 +109,13 @@
 	brnz,pn		%g1, 1b
 	 nop
  
-	sethi		%hi(p1275buf), %g2
-	or		%g2, %lo(p1275buf), %g2
-	ldx		[%g2 + 0x10], %l2
-	add		%l2, -(192 + 128), %sp
+	/* Get onto temporary stack which will be in the locked
+	 * kernel image.
+	 */
+	sethi		%hi(tramp_stack), %g1
+	or		%g1, %lo(tramp_stack), %g1
+	add		%g1, TRAMP_STACK_SIZE, %g1
+	sub		%g1, STACKFRAME_SZ + STACK_BIAS + 256, %sp
 	flushw
  
 	/* Setup the loop variables:
@@ -394,7 +397,6 @@
 	sllx		%g5, THREAD_SHIFT, %g5
 	sub		%g5, (STACKFRAME_SZ + STACK_BIAS), %g5
 	add		%g6, %g5, %sp
-	mov		0, %fp
  
 	rdpr		%pstate, %o1
 	or		%o1, PSTATE_IE, %o1
@@ -160,6 +160,36 @@
 	return 1;
 }
  
+int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
+			  struct page **pages)
+{
+	struct mm_struct *mm = current->mm;
+	unsigned long addr, len, end;
+	unsigned long next, flags;
+	pgd_t *pgdp;
+	int nr = 0;
+
+	start &= PAGE_MASK;
+	addr = start;
+	len = (unsigned long) nr_pages << PAGE_SHIFT;
+	end = start + len;
+
+	local_irq_save(flags);
+	pgdp = pgd_offset(mm, addr);
+	do {
+		pgd_t pgd = *pgdp;
+
+		next = pgd_addr_end(addr, end);
+		if (pgd_none(pgd))
+			break;
+		if (!gup_pud_range(pgd, addr, next, write, pages, &nr))
+			break;
+	} while (pgdp++, addr = next, addr != end);
+	local_irq_restore(flags);
+
+	return nr;
+}
+
 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
 			struct page **pages)
 {
@@ -11,11 +11,10 @@
 	.text
 	.globl	prom_cif_direct
 prom_cif_direct:
+	save	%sp, -192, %sp
 	sethi	%hi(p1275buf), %o1
 	or	%o1, %lo(p1275buf), %o1
-	ldx	[%o1 + 0x0010], %o2	! prom_cif_stack
-	save	%o2, -192, %sp
-	ldx	[%i1 + 0x0008], %l2	! prom_cif_handler
+	ldx	[%o1 + 0x0008], %l2	! prom_cif_handler
 	mov	%g4, %l0
 	mov	%g5, %l1
 	mov	%g6, %l3
@@ -26,13 +26,13 @@
  * It gets passed the pointer to the PROM vector.
  */
  
-extern void prom_cif_init(void *, void *);
+extern void prom_cif_init(void *);
  
-void __init prom_init(void *cif_handler, void *cif_stack)
+void __init prom_init(void *cif_handler)
 {
 	phandle node;
  
-	prom_cif_init(cif_handler, cif_stack);
+	prom_cif_init(cif_handler);
  
 	prom_chosen_node = prom_finddevice(prom_chosen_path);
 	if (!prom_chosen_node || (s32)prom_chosen_node == -1)
@@ -20,7 +20,6 @@
 struct {
 	long prom_callback;			/* 0x00 */
 	void (*prom_cif_handler)(long *);	/* 0x08 */
-	unsigned long prom_cif_stack;		/* 0x10 */
 } p1275buf;
  
 extern void prom_world(int);
@@ -52,6 +51,5 @@
 void prom_cif_init(void *cif_handler, void *cif_stack)
 {
 	p1275buf.prom_cif_handler = (void (*)(long *))cif_handler;
-	p1275buf.prom_cif_stack = (unsigned long)cif_stack;
 }
...	...	@@ -62,7 +62,8 @@
62	62	/* You must call prom_init() before using any of the library services,
63	63	* preferably as early as possible. Pass it the romvec pointer.
64	64	*/
65		-void prom_init(void cif_handler, void cif_stack);
	65	+void prom_init(void *cif_handler);
	66	+void prom_init_report(void);
66	67
67	68	/* Boot argument acquisition, returns the boot command line string. */
68	69	char *prom_getbootargs(void);
...	...	@@ -48,6 +48,8 @@
48	48	#endif
49	49
50	50	#ifdef CONFIG_SPARC64
	51	+void __init start_early_boot(void);
	52	+
51	53	/* unaligned_64.c */
52	54	int handle_ldf_stq(u32 insn, struct pt_regs *regs);
53	55	void handle_ld_nf(u32 insn, struct pt_regs *regs);
...	...	@@ -65,13 +65,10 @@
65	65	extern struct pause_patch_entry __pause_3insn_patch,
66	66	__pause_3insn_patch_end;
67	67
68		-void __init per_cpu_patch(void);
69	68	void sun4v_patch_1insn_range(struct sun4v_1insn_patch_entry *,
70	69	struct sun4v_1insn_patch_entry *);
71	70	void sun4v_patch_2insn_range(struct sun4v_2insn_patch_entry *,
72	71	struct sun4v_2insn_patch_entry *);
73		-void __init sun4v_patch(void);
74		-void __init boot_cpu_id_too_large(int cpu);
75	72	extern unsigned int dcache_parity_tl1_occurred;
76	73	extern unsigned int icache_parity_tl1_occurred;
77	74
...	...	@@ -672,14 +672,12 @@
672	672	sethi %hi(init_thread_union), %g6
673	673	or %g6, %lo(init_thread_union), %g6
674	674	ldx [%g6 + TI_TASK], %g4
675		- mov %sp, %l6
676	675
677	676	wr %g0, ASI_P, %asi
678	677	mov 1, %g1
679	678	sllx %g1, THREAD_SHIFT, %g1
680	679	sub %g1, (STACKFRAME_SZ + STACK_BIAS), %g1
681	680	add %g6, %g1, %sp
682		- mov 0, %fp
683	681
684	682	/* Set per-cpu pointer initially to zero, this makes
685	683	* the boot-cpu use the in-kernel-image per-cpu areas
686	684
687	685
...	...	@@ -706,44 +704,14 @@
706	704	nop
707	705	#endif
708	706
709		- mov %l6, %o1 ! OpenPROM stack
710	707	call prom_init
711	708	mov %l7, %o0 ! OpenPROM cif handler
712	709
713		- /* Initialize current_thread_info()->cpu as early as possible.
714		- * In order to do that accurately we have to patch up the get_cpuid()
715		- * assembler sequences. And that, in turn, requires that we know
716		- * if we are on a Starfire box or not. While we're here, patch up
717		- * the sun4v sequences as well.
	710	+ /* To create a one-register-window buffer between the kernel's
	711	+ * initial stack and the last stack frame we use from the firmware,
	712	+ * do the rest of the boot from a C helper function.
718	713	*/
719		- call check_if_starfire
720		- nop
721		- call per_cpu_patch
722		- nop
723		- call sun4v_patch
724		- nop
725		-
726		-#ifdef CONFIG_SMP
727		- call hard_smp_processor_id
728		- nop
729		- cmp %o0, NR_CPUS
730		- blu,pt %xcc, 1f
731		- nop
732		- call boot_cpu_id_too_large
733		- nop
734		- /* Not reached... */
735		-
736		-1:
737		-#else
738		- mov 0, %o0
739		-#endif
740		- sth %o0, [%g6 + TI_CPU]
741		-
742		- call prom_init_report
743		- nop
744		-
745		- /* Off we go.... */
746		- call start_kernel
	714	+ call start_early_boot
747	715	nop
748	716	/* Not reached... */
749	717
...	...	@@ -109,7 +109,6 @@
109	109	sllx %g5, THREAD_SHIFT, %g5
110	110	sub %g5, (STACKFRAME_SZ + STACK_BIAS), %g5
111	111	add %g6, %g5, %sp
112		- mov 0, %fp
113	112
114	113	call init_irqwork_curcpu
115	114	nop
...	...	@@ -30,6 +30,7 @@
30	30	#include <linux/cpu.h>
31	31	#include <linux/initrd.h>
32	32	#include <linux/module.h>
	33	+#include <linux/start_kernel.h>
33	34
34	35	#include <asm/io.h>
35	36	#include <asm/processor.h>
...	...	@@ -162,7 +163,7 @@
162	163
163	164	static struct pt_regs fake_swapper_regs = { { 0, }, 0, 0, 0, 0 };
164	165
165		-void __init per_cpu_patch(void)
	166	+static void __init per_cpu_patch(void)
166	167	{
167	168	struct cpuid_patch_entry *p;
168	169	unsigned long ver;
...	...	@@ -254,7 +255,7 @@
254	255	}
255	256	}
256	257
257		-void __init sun4v_patch(void)
	258	+static void __init sun4v_patch(void)
258	259	{
259	260	extern void sun4v_hvapi_init(void);
260	261
261	262
262	263
...	...	@@ -323,14 +324,25 @@
323	324	}
324	325	}
325	326
326		-#ifdef CONFIG_SMP
327		-void __init boot_cpu_id_too_large(int cpu)
	327	+void __init start_early_boot(void)
328	328	{
329		- prom_printf("Serious problem, boot cpu id (%d) >= NR_CPUS (%d)\n",
330		- cpu, NR_CPUS);
331		- prom_halt();
	329	+ int cpu;
	330	+
	331	+ check_if_starfire();
	332	+ per_cpu_patch();
	333	+ sun4v_patch();
	334	+
	335	+ cpu = hard_smp_processor_id();
	336	+ if (cpu >= NR_CPUS) {
	337	+ prom_printf("Serious problem, boot cpu id (%d) >= NR_CPUS (%d)\n",
	338	+ cpu, NR_CPUS);
	339	+ prom_halt();
	340	+ }
	341	+ current_thread_info()->cpu = cpu;
	342	+
	343	+ prom_init_report();
	344	+ start_kernel();
332	345	}
333		-#endif
334	346
335	347	/* On Ultra, we support all of the v8 capabilities. */
336	348	unsigned long sparc64_elf_hwcap = (HWCAP_SPARC_FLUSH \| HWCAP_SPARC_STBAR \|
...	...	@@ -109,10 +109,13 @@
109	109	brnz,pn %g1, 1b
110	110	nop
111	111
112		- sethi %hi(p1275buf), %g2
113		- or %g2, %lo(p1275buf), %g2
114		- ldx [%g2 + 0x10], %l2
115		- add %l2, -(192 + 128), %sp
	112	+ /* Get onto temporary stack which will be in the locked
	113	+ * kernel image.
	114	+ */
	115	+ sethi %hi(tramp_stack), %g1
	116	+ or %g1, %lo(tramp_stack), %g1
	117	+ add %g1, TRAMP_STACK_SIZE, %g1
	118	+ sub %g1, STACKFRAME_SZ + STACK_BIAS + 256, %sp
116	119	flushw
117	120
118	121	/* Setup the loop variables:
...	...	@@ -394,7 +397,6 @@
394	397	sllx %g5, THREAD_SHIFT, %g5
395	398	sub %g5, (STACKFRAME_SZ + STACK_BIAS), %g5
396	399	add %g6, %g5, %sp
397		- mov 0, %fp
398	400
399	401	rdpr %pstate, %o1
400	402	or %o1, PSTATE_IE, %o1
...	...	@@ -160,6 +160,36 @@
160	160	return 1;
161	161	}
162	162
	163	+int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
	164	+ struct page **pages)
	165	+{
	166	+ struct mm_struct *mm = current->mm;
	167	+ unsigned long addr, len, end;
	168	+ unsigned long next, flags;
	169	+ pgd_t *pgdp;
	170	+ int nr = 0;
	171	+
	172	+ start &= PAGE_MASK;
	173	+ addr = start;
	174	+ len = (unsigned long) nr_pages << PAGE_SHIFT;
	175	+ end = start + len;
	176	+
	177	+ local_irq_save(flags);
	178	+ pgdp = pgd_offset(mm, addr);
	179	+ do {
	180	+ pgd_t pgd = *pgdp;
	181	+
	182	+ next = pgd_addr_end(addr, end);
	183	+ if (pgd_none(pgd))
	184	+ break;
	185	+ if (!gup_pud_range(pgd, addr, next, write, pages, &nr))
	186	+ break;
	187	+ } while (pgdp++, addr = next, addr != end);
	188	+ local_irq_restore(flags);
	189	+
	190	+ return nr;
	191	+}
	192	+
163	193	int get_user_pages_fast(unsigned long start, int nr_pages, int write,
164	194	struct page **pages)
165	195	{
...	...	@@ -11,11 +11,10 @@
11	11	.text
12	12	.globl prom_cif_direct
13	13	prom_cif_direct:
	14	+ save %sp, -192, %sp
14	15	sethi %hi(p1275buf), %o1
15	16	or %o1, %lo(p1275buf), %o1
16		- ldx [%o1 + 0x0010], %o2 ! prom_cif_stack
17		- save %o2, -192, %sp
18		- ldx [%i1 + 0x0008], %l2 ! prom_cif_handler
	17	+ ldx [%o1 + 0x0008], %l2 ! prom_cif_handler
19	18	mov %g4, %l0
20	19	mov %g5, %l1
21	20	mov %g6, %l3
...	...	@@ -26,13 +26,13 @@
26	26	* It gets passed the pointer to the PROM vector.
27	27	*/
28	28
29		-extern void prom_cif_init(void , void );
	29	+extern void prom_cif_init(void *);
30	30
31		-void __init prom_init(void cif_handler, void cif_stack)
	31	+void __init prom_init(void *cif_handler)
32	32	{
33	33	phandle node;
34	34
35		- prom_cif_init(cif_handler, cif_stack);
	35	+ prom_cif_init(cif_handler);
36	36
37	37	prom_chosen_node = prom_finddevice(prom_chosen_path);
38	38	if (!prom_chosen_node \|\| (s32)prom_chosen_node == -1)